Figure 4. Floor texture simulation, results.

(a) During the inverse task the parameters chosen during the exploratory phase were replayed in a random order; subject had to report the floor type relying on the tactile feedback only (chance level = 0.42). (b) Patients’ accuracy per floor type (multiple comparisons, N = 14) and for trials with or without floor. (c) Mean answer over all patients and responses to Q1 (asked prior to the experiment): ‘I remember the sensation of my feet on X’ (X: SAT,GRT or PST, ‘I strongly disagree’ −2 to ‘I strongly agree’ +2). (d) Detail of patients responses for Q1. (e) Subjects’ responses to question ‘I had during the experiment the sensation of walking on X’ after a V only session (patients observed the avatar walking) and a Vision + tactile session (tactile feedback on stance) (*for P < 0.05, Wilcoxon test). (f) Detail of patients’ answers to question Q2. (g) Principal tactile factors of the experiment, ground types are classified (using knn classification) using each one of the four factors. Mean classification accuracy over all sessions for four factors (multiple-comparison test, N = 14, *: P < 0.05). (h) Mean and standard deviation for principal factors: Amplitude of proximal Vibration (PV) and Stimulation Timing (ST) for all sessions. Color convention as follows: red for SAT, green for GRT and blue for PST. (i) Values from panel h are joined following structure similarities (e.g. sessions where SAT was on top left corner and PST on the lower right).